Re: Friendliness and blank-slate goal bootstrap

Date: Mon Oct 06 2003 - 18:09:17 MDT

In a message dated 10/6/2003 9:36:58 AM Eastern Standard Time, writes:

> Of course, in perspective 2 (the scientific perspective), one never answers
> the question whether the entity "*really* experiences qualia" because the
> answer to that question lies in the experiential rather than scientific
> domain.
> -- Ben Goertzel

All this talk of qualia reminds me of Justice Potter Stewart's famous line:
"I can't define obscenity, but I know it when I see it."
In other words, he would filter the allegedly obscene writings, etc. through
some template in his own mind, and either it would fit his template or not.
Even among humans the various mental templates we encounter are wildly
divergent, and neuroscientists can often locate (with PET scans and other tools)
where the differences in activity are from one brain to the next, so that a
sociopathic murderer, who can fool strangers on the street with his pseudo-friendly
manner, stands revealed when the wrong parts of his brain activate (or don't
activate) while under the penetrating gaze of the PET scan. Does this mean Ted
Bundy had different qualia, or even none as we would agree to the meaning of
the word? I'm not sure the word is even useful.
The sociopath example implies that the failure of some separate inhibitory
module (analogous to the forebrain) might turn an AI from friendly to
unfriendly, if such a module existed. It is now known that reckless, thoughtless
adolescent behavior is due to the fact that this region does not mature until near
the twentieth birthday. Could this be a risk with a young AI? "I know I
vaporized Utah, but I swear I won't do it again!"
Tom Buckner

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:42 MDT